Conditions

Conditions Values
N_sample 100, 300, 500
N_Items 10, 50, 100
Omega 0, 0.5
r^2 of Cov on LV 0.1, 0.5
  • Total = 36 conditions x (four types of priors + ML)

Results

  • Pirors
    • alldiffuse: Diffuse prior for all parameters (Uniform)
    • measureinfo: Informative prior for only measurement params (Normal, LogN)
    • allinfo: Informative for all params
    • matched: matched priors to the population value with small variance (like N(0.5, 0.1) when the population value is 0.5)
  • 5 replications per condition
  • The sample size are only considered

Measurement part

  • ML saw some extreme outliers, so the range of y is limited between -1 and 1 as the second plot.

  • Number of items means the number of items for the factor model (2pl).

  • Omega is the effect from the latent factor on the outcome (y).

  • 0 means no relationship between the latent factor and y.

  • Power_on_latent is the proportion of latent variance explained by the two covariates.

  • 0.1 means only 10% variance explained by two covariates.

By sample size and number of items

By sample size and omega

By sample size and power_on_latent

Structural part

By sample size and number of items

By sample size and omega

By sample size and power_on_latent